Research map / Detection and segmentation / CNN features
Learning To Discover Reflection Symmetry via Polar Matching Convolution
Ahyun Seo, Woohyeon Shim, Minsu Cho
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| add polar self-similarity descriptor to polar region descriptor (ablation) | PMCNet · SDRW test | max F1: 65.5 -> 68.3 | Improved |
| add polar self-similarity descriptor to polar region descriptor (ablation) | PMCNet · LDRS test | max F1: 35.3 -> 35.9 | Improved |
| add polar region descriptor to self-similarity descriptor (ablation) | PMCNet · LDRS test | max F1: 33.7 -> 35.9 | Improved |
| make kernel axis-aware (on learnable kernel) (ablation) | PMCNet · SDRW test | max F1: 67.5 -> 68.8 | Improved |
| make kernel learnable (on axis-aware kernel) (ablation) | PMCNet · LDRS test | max F1: 34.6 -> 37.3 | Improved |
| add base feature F to decoder input (ablation) | PMCNet · SDRW test | max F1: 62.7 -> 68.8 | Improved |
| add polar matching scores to decoder input (ablation) | PMCNet · SDRW test | max F1: 66.8 -> 68.8 | Improved |
| add ImageNet pre-training (ablation) | PMCNet · SDRW test | max F1: 47.8 -> 68.8 | Improved |
| fine-tune backbone (ablation) | PMCNet · LDRS test | max F1: 32.7 -> 37.3 | Improved |
| add synthetic LDRS-synth data to real data (ablation) | PMCNet · SDRW test | max F1: 61.6 -> 68.8 | Improved |
| add synthetic LDRS-synth data to real data (ablation) | PMCNet · LDRS test | max F1: 34.8 -> 37.3 | Improved |
| add real data to LDRS-synth (ablation) | PMCNet · LDRS test | max F1: 14.2 -> 37.3 | Improved |
Closest papers by meaning
- Symmetry Strikes Back: From Single-Image Symmetry Detection to 3D Generation (CVPR 2025 · similarity 0.87)
- Symmetry Induces Structure and Constraint of Learning (ICML 2024 · similarity 0.85)
- RGB-to-Polarization Estimation: A New Task and Benchmark Study (NeurIPS 2025 · similarity 0.85)
- Leveraging 3D Geometric Priors in 2D Rotation Symmetry Detection (CVPR 2025 · similarity 0.85)
- Symmetry Discovery Beyond Affine Transformations (NeurIPS 2024 · similarity 0.85)
- Discovering Symmetry Breaking in Physical Systems with Relaxed Group Convolution (ICML 2024 · similarity 0.85)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
